FRS Topologies

FRS supports arbitrary topologies subject to requirement of the replica set type.

FRS does not rely on broadcast technology and can provide fault tolerant distribution via multiple connection paths between members. If a given replica member is unavailable, the data will flow via a different route assuming one exists. Dampening logic prevents a data file from being sent more than once to any given member.

Replicating data between sites that are connected with slow WAN links requires careful planning of the topology and schedule. If the sites have a high-bandwidth connection, but business-critical databases and other applications use that connection as well, you will need to schedule replication so that it does not consume bandwidth needed for other uses.

A key part of topology definition involves choosing a trade-off between redundancy and scalability. Topologies with more redundant paths offer greater fault tolerance by providing the largest number of alternate routes for data transfer, but at the same time this adds administrative overhead (more connections to monitor and support) and potential liability (a single offline downstream partner now affects additional upstream partners).

The full mesh topology in which every node has a connection to every other is the most extreme case: it provides the most redundancy, but the least scalability. By contrast, the single-hub-and-spoke topology provides the least redundancy, but provides better scalability. Consequently, the best topology for a given scenario typically lies in a continuum between the extremes of full mesh and single-hub-and-spoke, and balances scalability and redundancy.

In the majority of cases (especially large ones) the optimal topology is often redundant hub-and-spoke with staggered schedules. In this arrangement, a small number of members (the hubs) have direct inbound and outbound connections with every other member (the spokes and the other hubs). There are no additional connections. The connections from the hubs to the spokes have staggered schedules so that the hubs do not get overloaded by replicating files to all partners at once. The hubs are well-connected to each other and typically have constant (or very frequent) schedules.

A case that is common within sites is the ring topology. In this case, all members are connected as if they were points on a circle, and each member has inbound and outbound connections with only its immediate neighbors.

Configuring the desired replication topology (DFS case)

Administrators should use the updated Distributed File System snap-in the Windows Server 2003 Administration Tools Pack. You can install this pack on computers running Windows XP with Service Pack 1 (SP1) or later and create FRS schedules and topologies on remote servers running Windows 2000. This snap-in offers a number of important benefits over the version of the tool that shipped in Windows 2000:

For more information, see article 304718, "Administering Windows 2000-Based and Windows Server 2003-Based Computers Using Windows XP Professional-Based Clients" at http://support.microsoft.com/?id=304718.

Configuring the desired replication topology (SYSVOL case)

The techniques and tools for configuring SYSVOL replication sets is described in length, along with tools and scripts, in the Windows 2000 Branch Office Deployment Guide available at http://go.microsoft.com/fwlink/?LinkId=16718.

It is essential that administrators with large configurations (over 100 sites) review the Branch Office Deployment Guide (especially the topology section) to decide if a custom, scripted topology is required. Even for smaller configurations, this is a very important and useful document and tool set.